Using Homemovers to Identify Local Campaign Effects

Abstract

Canonical models of voter turnout suggest that electoral competitiveness drives participation in elections. Yet while this relationship holds true in aggregate analyses, the individual-level evidence is mixed. While the limitations of using cross-sectional data to approximate causal effects are well known, this paper demonstrates that even longitudinal designs typically estimate treatment effects that are significantly unrepresentative of the wider population. This study develops a new identification strategy to overcome these shortcomings. Focusing on those who move house between elections, I show that the political characteristics of homemovers’ destinations are conditionally independent to potential outcomes. This allows for identification of the conditional average treatment effect of local context on voter turnout. I apply this framework to the 2017 and 2019 general elections in the United Kingdom, and show that moving to a more competitive constituency significantly increases campaign contact, but that this has only mixed effects on the probability of turning out to vote. The paper offers a methodological contribution to the study of political behaviour, while adding to the literature on campaign effects and participation.